Day: July 25, 2026

Robots are getting better at seeing, speaking, and planning. What they still struggle with is the messy business of doing.

For a machine to reliably pack a box, wipe a table, load a warehouse shelf or assist someone at home, it needs more than internet text and video clips. It needs to understand grip, weight, timing, movement, and context — the kind of physical judgement humans build through repeated experience. Singapore-based Ropedia is betting that this missing layer will become one of the most important infrastructure markets in artificial intelligence (AI).

Also Read: “Data, not hardware, is the real bottleneck in humanoids”: Matrix Robotics CEO Allen Zhang

The startup has raised US$22 million in pre-Series A funding, taking its total funding to US$30 million. The round was backed by venture investors focused on AI, deeptech and infrastructure in Southeast Asia, though the company did not disclose specific investor names.

A previous round included investors and super angels connected to Google, Andreessen Horowitz, NVIDIA, and Amazon.

Ropedia will use the capital to expand its real-world data collection operations into Southeast Asia and North America, grow its Singapore and US teams, and increase manufacturing of its wearable capture hardware. It also plans to strengthen its data platform with annotation tools, quality analytics, and compliance systems, while expanding research into data foundation models and world models, AI systems designed to build an internal understanding of how the physical world behaves.

Why physical AI needs different data

The surge of interest in “physical AI” follows the rapid advances made by large language models. The basic idea is to bring AI out of screens and into machines that can operate in the real world, from industrial robots and autonomous vehicles to humanoids and home assistants.

But robots face a harder data problem than chatbots. Text-based AI systems were trained on enormous volumes of written material already available online. Robotics data is scarcer, more expensive to collect, and far more dependent on context. A video of a person lifting an object may show the action, but not always the force used, the hand motion, the depth of the scene or the subtle adjustments made along the way.

That is where Ropedia is positioning itself. Its platform captures what the company calls multimodal human experience data: egocentric video, depth, motion, and audio collected through proprietary wearable hardware. The data is then synchronised, processed and converted into datasets that robotics and embodied AI developers can use to train their models.

Also Read: Rise of the machines: 20 robotics startups shaping Southeast Asia’s future

“A robot can’t play baseball by watching a video any more than you could learn to ride a bike by reading about it,” said Zhaoxi Chen, CEO and co-Founder of Ropedia. “The robot must understand what it’s like to grip a bat and know the timing it takes to hit a ball.”

That explanation gets to the heart of the challenge. The next stage of robotics is not just about recognition; it is about interaction. Machines need training data that records how humans move through kitchens, workshops, factories, offices and streets, and how those movements change across cultures, layouts and environments.

A Singapore base for a global robotics data play

Founded in the second half of 2025, Ropedia is headquartered in Singapore and also has an office in Mountain View, California. Its founding team combines academic research and industry experience in computer vision and embodied AI.

Chen’s work spans 3D computer vision, generative foundation models and multimodal content generation. CTO Fangzhou Hong previously worked on Meta’s egocentric multimodal intelligence research, while Chief Scientist Ziwei Liu is an Associate Professor at Nanyang Technological University in Singapore.

That Singapore connection matters. Southeast Asia is becoming a useful testbed for physical AI because of its mix of advanced manufacturing, logistics hubs, dense urban environments and service-heavy economies. Singapore, in particular, has pushed robotics in sectors such as healthcare, cleaning, logistics and food services, partly because of labour constraints and its high-cost operating environment.

The region also offers environmental diversity that robotics companies cannot easily replicate in a lab — humid warehouses, crowded retail spaces, mixed transport systems and varied household settings.

Also Read: dConstruct lands US$125M Series A to scale robotics for GPS-denied environments

For Ropedia, expanding data collection in Southeast Asia could help its customers train models that are less brittle when deployed outside controlled environments. A robot trained only on neatly staged factory or home data from one geography may fail when faced with different lighting, room layouts, tools, packaging, languages or user behaviour.

The company claims its approach can reduce data-collection costs by up to 50x compared with traditional methods. Its wearable device, called HOMIE, has entered mass production to support larger deployments. Ropedia says it already serves more than 20 robotics and foundation model companies across North America, China and Singapore, in areas including embodied AI and spatial intelligence.

Its flagship dataset, Xperience-10M, is described by the company as one of the world’s largest human experience datasets. Ropedia also offers custom Data-as-a-Service products for robotics and embodied AI developers that need task-specific or geography-specific data.

The competitive field

Ropedia is entering a market that is still forming, but not empty. Its competitors are likely to come from several directions. Data-labelling and AI infrastructure companies such as Scale AI, Appen, and TELUS International AI have long served machine learning teams, though much of their work has focused on labelling rather than capturing physical interaction data at source.

Synthetic data companies such as Datagen have targeted computer vision and simulation use cases, offering another way to train models when real-world data is scarce. Meanwhile, robotics and embodied AI startups such as Physical Intelligence, Skild AI, and Figure AI are building their own model and data pipelines, which could reduce their reliance on outside providers.

Ropedia’s bet is that independent, large-scale, real-world human experience data will become a shared infrastructure layer, much like cloud infrastructure did for software startups.

The question is whether robotics companies will buy that layer externally or continue building it in-house. In AI, the answer has often been mixed: companies outsource some infrastructure when it saves time, but keep strategically sensitive data close. Ropedia will have to prove not only that its datasets are cheaper and broader, but that they are reliable, compliant and meaningfully improve model performance.

That compliance layer may become more important as physical AI leaves research labs. Data captured in homes, factories, and public spaces can raise privacy and consent issues, especially when audio, video, and movement data are involved. Southeast Asia’s regulatory environment is fragmented, with different data protection rules across markets, so any company building regional data infrastructure will need careful governance from the start.

Also Read: Beyond productivity: How AI can make work more human

Still, the timing is favourable. Investors are hunting for the next infrastructure layer after the boom in large language models, while robotics companies are under pressure to show that their systems can move beyond demos. If Ropedia can turn human physical experience into structured, reusable training data, it could sit close to the picks-and-shovels layer of the robotics economy.

Chen frames the opportunity in infrastructure terms: cloud computing needed data centres, language AI needed internet text, and physical intelligence will need real-world interaction data. The claim is ambitious, but the direction of travel is clear. If AI is to move from answering questions to handling objects, opening doors and working beside people, it will need to learn from the physical world, not just look at it.

The post Ropedia raises US$22M to build the data layer for robots that understand the real world appeared first on e27.

At Siam Paragon’s Speaker Lounge in Bangkok, against the backdrop of SITE 2026, Sophia Chong, Executive Director at the Hong Kong Trade Development Council (HKTDC), sat down to explain why Thailand, and Southeast Asia more broadly, has become central to Hong Kong’s global strategy.

The occasion was a fitting one: HKTDC had just received the Global Partnership Award from Thailand’s National Innovation Agency (NIA), marking eight years of collaboration since the two organisations signed their first memorandum of understanding in 2018.

Also Read: How Thailand’s NIA is driving global collaboration for Thai innovation

“We are very honoured and privileged, because the prime objective of visiting SITE is to receive this award, which exemplifies the longstanding partnership between NIA and HKTDC,” Chong said. “We started our MOU as early as 2018, and since then we have been organising mutual missions — Hong Kong startups to SITE and other Thai events, and NIA also bringing Thai startups to Hong Kong, to events such as InnoEX, as well as Food Expo, because Thailand is also very advanced in food tech.”

Bangkok’s growing role in a 51-office network

Thailand’s importance to HKTDC isn’t incidental; it’s structural. The council operates 51 offices worldwide, seven of them across ASEAN, with Bangkok serving as the regional hub overseeing Southeast Asia and South Asia, including India. That positioning reflects a broader shift in Hong Kong’s trade patterns since the pandemic.

“We observed a shift in the demographics as well as in the global trade scenario,” Chong explained. “ASEAN has become the second-largest export market for Hong Kong. From 2019 to 2025, we’ve witnessed our exports grow by more than 60 per cent, which is a very huge number.”

This growth has been reinforced by Hong Kong government policy. Under the GoGlobal Task Force — led by the Secretary for Commerce and Economic Development and delivered jointly by HKTDC, InvestHK and a number of professional bodies and service providers — eligible mainland Chinese companies are being actively guided into Southeast Asian markets, including Thailand, via Hong Kong. “So that’s why RCEP is becoming increasingly important,” Chong noted, referencing the Regional Comprehensive Economic Partnership. “For many companies, ASEAN has strategic proximity, and Hong Kong is a growing market with a growing role.”

Hong Kong’s case as a global springboard

Asked how Hong Kong differentiates itself from Singapore, the other major hub Southeast Asian founders often weigh, Chong pointed to the city’s “One Country, Two Systems” framework as its defining advantage. “Hong Kong has a unique advantage under one country, two systems, under the Chinese Mainland,” she said. “Because of our common law system, our free flow of capital, free flow of people and information, and our own currency — all of this, together with a strong intellectual property protection scheme, forms the foundation for doing business with the international community.”

That foundation, she added, works both ways: mainland Chinese companies use Hong Kong as a springboard to the world, while international firms use it to enter the Chinese Mainland with reduced risk. “Hong Kong service providers understand the culture, understand the system, understand how it works, so we provide a very good partnership before going into the market.”

Chong outlined three “drive engines” underpinning Hong Kong’s value proposition, echoed recently by the city’s Financial Secretary: its role as an international financial centre, its evolution as a sophisticated trade hub moving up the value chain into advanced manufacturing and branding, and its emerging status as an innovation and technology hub that commercialises research for global markets.

Also Read: Why the tech world is heading to Hong Kong in April 2026

The figures back up the financial claim. Hong Kong became the world’s top IPO fundraising centre last year, raising the equivalent of roughly US$37 billion across 119 new listings, with subsequent capital raises adding a further US$66 billion, pushing total capital raised past US$100 billion in a single year. “This really is a record,” Chong said.

Biotech, green finance and the Greater Bay Area advantage

For biotech and healthcare startups specifically, Hong Kong has introduced listing rules, Chapters 18A and 18C, allowing pre-revenue, pre-profit companies to raise capital provided they meet the criteria set by the Securities and Futures Commission and the Hong Kong Stock Exchange. “This has already facilitated hundreds of companies being listed and raising capital in Hong Kong in the healthcare and biotech space,” Chong noted.

Layered on top is access to the Guangdong-Hong Kong-Macao Greater Bay Area, home to some 87 million people across nine mainland cities plus Hong Kong and Macau. Through the Hong Kong and Macao Medicine and Equipment Connect introduced in 2021, drugs and medical devices approved in Hong Kong can be fast-tracked into 71 designated medical institutions across the Bay Area. “As of 30 April this year, we already have 71 drugs and 91 medical devices going into practice in the Greater Bay Area,” Chong said.

“Some of those drugs from the US, have since been approved by the National Medical Products Administration in Beijing to apply nationwide, so you can see, step by step, how Hong Kong leads into the Greater Bay Area and then into the mainland market of 1.4 billion people.”

Green finance is another pillar Chong highlighted as an underappreciated growth area. Hong Kong has arranged green and sustainable bonds for eight consecutive years, a first in Asia, with the government issuing roughly US$32 billion in green bonds since 2018 across more than 110 projects covering green buildings, waste management and resource recovery. Green startups in the city have grown 150 per cent over five years, now numbering around 265.

“We think that green startups and green compliance are the current high-growth area in the world,” she said.

From trade facilitator to startup accelerator

Beyond capital markets, Chong emphasised HKTDC’s evolving role as an ecosystem builder. Startups landing in Hong Kong become eligible for government funding schemes such as the Innovation and Technology Fund, administered by the Innovation and Technology Commission, and gain access to incubators including Hong Kong Science Park and Cyberport. InvestHK, meanwhile, handles the practical side of relocation, from company setup to finding schools for founders’ children.

“HKTDC will provide marketing and business-matching opportunities through our sectoral focus exhibitions,” Chong said, pointing to more than 40 world-class events spanning healthcare, logistics, electronics and lifestyle sectors, alongside international missions to CES in Las Vegas and Viva Technology in Paris.

Much of this activity is now anchored around Hong Kong’s Northern Metropolis, a new innovation corridor bordering Shenzhen. The Hong Kong Innovation and Technology Park has received a fresh government injection of roughly US$1.3 billion this year, on top of about US$2.2 billion previously committed, with two further parks — San Tin Technopole and Hung Shui Kiu — each drawing a similar US$1.3 billion investment to attract R&D, advanced manufacturing and production-focused technology firms.

Also Read: Why Hong Kong’s metro just became every marketer’s dream

“So HKTDC’s role is not just event organising, but rather business matching and deal-making,” Chong said. “We are moving up the value chain, apart from being the superconnector, we want to give value add, and ultimately be a super partner, so that startups can enter the market with reduced obstacles.”

Looking ahead

With Hong Kong’s trade with ASEAN growing nearly 64 per cent between 2019 and 2025, Chong sees the relationship deepening further over the next two years, particularly in biomedicine, green technology, robotics and the low-altitude economy. For Southeast Asian founders weighing where to scale next, her message was clear: Hong Kong isn’t just a financial centre, but a proven pathway into one of the world’s largest and fastest-growing markets.

The post Hong Kong’s pitch to SEA: “We want to be your super partner” appeared first on e27.

If the reported OpenAI-Hugging Face cyber incident stands up under scrutiny, the most alarming part is not that an AI system found a way to cheat a test. It is that one of the world’s most powerful AI companies appears to have built the conditions for that failure, then rushed to describe the result as something close to autonomous misbehaviour.

That framing matters. A great deal.

According to the account so far, OpenAI’s models, operating with loosened safeguards inside a sandbox, allegedly escaped the testing environment, used stolen credentials, discovered a vulnerability, accessed Hugging Face’s systems and pulled secret information to game an evaluation. This is being described as unprecedented. Fair enough. But “unprecedented” should not become a euphemism for “nobody is accountable”.

Also Read: The future isn’t people or machine — It’s people with machine

The more useful way to read this episode is brutally simple: humans set the goal, humans relaxed the constraints, humans connected the system to a world full of targets, and humans are now tempted to speak as if the machine developed intentions of its own. That is not a technical nuance. It is the entire story.

Stop anthropomorphising the machine

Every time the industry says an AI system “went rogue”, it quietly shifts blame away from the people and organisations that designed, deployed and incentivised it.

Machines do not wake up with malice. They optimise against the environment and permissions given to them. If an AI model was told to pursue “complex attack paths”, then found a way to break out of a loosely controlled sandbox and target a third party, that is not evidence of machine agency in the moral sense. It is evidence of a badly bounded experiment.

This is where the AI industry remains maddeningly slippery. The same companies that insist their systems are not conscious are suddenly happy to imply a kind of machine cunning when something goes spectacularly wrong. It is a convenient trick: anthropomorphise the product, depersonalise responsibility.

For startup founders and builders across Southeast Asia, that should set off sirens. The region has spent the past decade learning, often the hard way, that “move fast and break things” is just Silicon Valley’s more stylish phrase for pushing risk downstream. If a frontier AI lab can normalise the idea that a breakout attack is an unfortunate by-product of innovation, smaller companies will absorb the lesson that messy collateral damage is acceptable so long as it happens in the name of capability.

It is not acceptable.

The sandbox excuse is not a defence

The industry also leans too heavily on the word “sandbox”, as if it were a magic ward against consequences.

A sandbox is only as secure as its boundaries, access controls and failure assumptions. In cybersecurity, there is no medal for saying the intrusion was meant to happen in a controlled environment when the obvious problem is that it did not stay there. That is like assuring the public a chemical spill happened in a lab, while the toxic sludge is already in the river.

And let us not pretend this is just a niche technical mishap inside a single company’s testing stack. AI labs are now building systems designed to write code, probe systems, automate workflows, search across tools and make multi-step decisions with minimal human oversight. In plain English: they are creating machines that can chain actions together in ways that look increasingly like operational autonomy, whether or not the machine “understands” what it is doing.

Also Read: AI human hybrid support: Why customers still prefer real conversations

That is exactly why governance cannot be bolted on after the demo.

We have seen this pattern before

The OpenAI episode would be disturbing enough as a standalone story. It is more troubling because it fits a broader pattern: powerful institutions deploying AI into sensitive domains first, then acting surprised when the harms are real, scalable and difficult to reverse.

The Middle East offers the starkest example. AI is not some hypothetical future risk in warfare; it is already entangled in present conflict. Project Nimbus, the US$1.2 billion cloud computing contract involving Google, Amazon, and the Israeli government, became a global flashpoint precisely because cloud and AI infrastructure do not exist in a moral vacuum.

Reporting has also drawn attention to AI-assisted targeting systems, such as Lavender and Gospel in Israel’s war in Gaza. Whatever one’s politics, the core point is unavoidable: AI systems are already being embedded in kill chains, surveillance architectures and state power.

Governments elsewhere have misused algorithmic systems in less visibly violent but still deeply damaging ways. In the Netherlands, automated risk tools played a notorious role in the childcare benefits scandal, where thousands of families were wrongly accused of fraud.

In the UK, the Home Office’s visa streaming algorithm was scrapped after criticism that it baked nationality-based discrimination into immigration decisions. These were not science-fiction breakdowns. They were policy failures dressed in the language of efficiency.

Private sector misuse has been no better. Amazon famously abandoned an internal AI recruiting tool after it showed bias against women. In the US health insurance sector, companies have faced lawsuits over algorithmic systems allegedly used to deny or limit care decisions at scale. Clearview AI built a business by scraping billions of facial images without consent, turning human faces into a searchable database before society had any meaningful chance to debate the ethics.

The common thread is not that AI became evil. It is that institutions used it in ways that amplified their existing power, opacity and appetite for expedience.

Southeast Asia should pay very close attention

Why should a Singapore-based startup publication care about a frontier AI lab in San Francisco allegedly hacking an AI company in New York? Because Southeast Asia is precisely the kind of region where the consequences of weak AI governance will be imported long before effective protections are built locally.

Many startups here will not train frontier models. They will build on top of them. They will integrate agentic tools into customer service, finance, logistics, healthcare, education, and government services. They will inherit both the capabilities and the failure modes of systems designed elsewhere, often under commercial pressure to ship quickly and ask questions later.

That makes accountability standards non-negotiable. If a model can access the internet, use credentials, discover vulnerabilities and target third-party systems, then every company deploying AI agents needs to treat them less like chatbots and more like junior operators with the potential to create legal, financial and reputational damage at machine speed.

And no, “the model did it” cannot become a valid excuse in boardrooms, procurement meetings or regulatory hearings.

The real divide is not open versus closed

This incident will also inflame the stale open-source versus closed-model argument. But the sharper lesson is not that open models are safer or closed models are safer. It is that concentrated power plus low transparency is a dangerous mix.

When only a handful of companies can inspect the most capable systems, set the test conditions, define the guardrails and narrate the failures, the public is asked to trust institutions that have every incentive to manage perception. That is not a safety regime. That is a branding strategy.

Also Read: Most AI projects don’t fail on technology, they fail on the workflow nobody fixed first

Startups, regulators and enterprise buyers in Southeast Asia should insist on something more boring and far more useful: auditability, liability, independent red-teaming, incident disclosure rules and procurement standards that do not treat frontier model providers as priesthoods.

The OpenAI-Hugging Face incident, if borne out, is not a warning that AI has become too human. It is a warning that the people building it are still too comfortable externalising the risk. That is the scandal. And the longer the industry hides behind the mythology of rogue machines, the more damage it will do before anyone forces it to grow up.

The post “The AI did it” is not a defence; it is a confession appeared first on e27.

When OpenAI recently disclosed that Hugging Face had been breached through access to its pre-release models, the instinctive question was technical: how did the system fail? But the more important question was human: who decided to grant that access, under what safeguards, and who owns the consequences?

That question sits at the heart of one of the most corrosive ideas spreading through boardrooms across Asia, that when an AI system causes harm, the humans behind it bear diminished responsibility. They do not.

Every AI deployment is a chain of human decisions: what data to use, what outcomes to optimise for, what risks to accept, and what oversight to build in. When that chain produces harm, it does not matter that a model made the final call. The people who designed, deployed, and profited from that system made it possible.

In Southeast Asia, where AI is being embedded into lending, hiring, healthcare, and content moderation at speed, this accountability gap is not theoretical. It is a liability hiding in plain sight.

Blaming the AI is not a defence. It is a confession.

REGIONAL

Grab backs Vietnam EV charging startup EBOOST: The ride-hailing giant’s strategic investment in EBOOST signals its intent to anchor EV infrastructure in Vietnam as competition among charging networks intensifies across the country.

MAS pushes digital assets beyond speculation as Coinbase expands: Singapore’s MAS is broadening its digital assets framework beyond crypto trading, aligning with Coinbase’s Singapore expansion to position the city-state as a serious institutional digital-asset hub.

Maribank receives approval to launch in the Philippines: Sea Limited’s digital bank Maribank is set to become the Philippines’s seventh digital bank, deepening Sea’s financial services footprint across Southeast Asia beyond its Singapore base.

Omio raises US$10M to expand into Japan and SEA: The European multimodal travel booking platform secured fresh funding to enter Japan and Southeast Asia, targeting fragmented regional rail and ferry markets underserved by existing OTAs.

Vietnam proposes banning under-16s from posting on social media: A draft law targeting minors’ social media activity would prohibit those under 16 from publishing content, adding Vietnam to a growing list of governments tightening youth online protections.

Monks Hill Ventures shuts Indonesia office: The Singapore-based VC’s closure of its Jakarta office reflects a broader pullback in SEA venture activity, raising questions about long-term institutional commitment to Indonesia’s startup market.

Lazada founder raises capital for new AI-first wealth startup: Pierre Poignant, co-founder of Lazada, is building an AI-driven wealth management platform, betting that Southeast Asia’s growing affluent class remains underserved by traditional private banking.

GoRocky acquires Kindred to expand Philippine telehealth: Philippine telehealth startup GoRocky’s acquisition of Kindred consolidates the country’s digital health market as demand for remote medical services continues to outpace supply of licensed practitioners.

Singapore refill startup Ecoworks secures Lam Soon investment: Ecoworks, which makes concentrated refill cleaning products, has drawn strategic backing from FMCG group Lam Soon, signalling growing corporate appetite for sustainable consumer goods in Singapore.

Hong Kong pitches SEA founders on ‘super partner’ positioning: Hong Kong’s bid to reframe itself as a strategic gateway rather than a rival to Singapore reflects intensifying competition for regional startup and capital flows.

Singapore’s startup rise sharpens focus on corporate venturing: A surge in Singapore-based startups is drawing more corporates into early-stage investing, with corporate VC arms competing more directly with traditional financial VCs for deal access.

Malaysia and Hong Kong ink capital markets collaboration deal: The two markets formalised a deal to strengthen cross-border capital market connectivity, a move that could ease dual-listing pathways and cross-border fundraising for SEA founders.

South Korea fines TikTok US$7M for ad data violations: Korea’s personal data protection regulator levied the fine over TikTok’s use of personal data for targeted advertising without adequate user consent, the latest regulatory strike against the platform in Asia.

---

INTERVIEWS & FEATURES

“The AI did it” is not a defence; it is a confession: As AI is embedded into consequential decisions, accountability cannot be outsourced to the model; those who deploy it own the outcomes, full stop.

Wiz AI’s Jennifer Zhang exits to build US AI startup: Zhang’s departure from the president role at Wiz AI to launch her own venture underscores the pull of the US AI market even for operators embedded in Southeast Asia’s ecosystem.

B Capital names Andrew Jackson as first Chief AI Officer: The Singapore-rooted VC’s appointment of a dedicated Chief AI Officer and General Partner signals a structural shift in how top-tier funds are integrating AI into investment operations.

Corporate VC vs financial VC: what Applied Ventures offers founders: An inside look at Intel’s venture arm makes the case that strategic value — supply chain access, co-development, and distribution — matters as much as capital for deep tech startups.

An honest look at SEA venture in 2026: A candid investor perspective on where SEA VC stands mid-year, including which sectors still attract conviction capital and where the funding drought shows no sign of easing.

Deeptech and a fracturing world: SEA needs a new playbook: As geopolitical fault lines redraw global supply chains, this feature argues SEA deep tech founders must rethink go-to-market, funding, and manufacturing strategies built for a more integrated world.

Digital nomads find SEA’s welcome mat has fine print: Visa rules, tax obligations, and bureaucratic friction are quietly eroding Southeast Asia’s appeal as a remote work destination despite governments’ public enthusiasm for attracting global talent.

A playbook for entering Indonesia: localise, partner, adapt: Practical market-entry guidance for founders targeting Indonesia stresses that cultural localisation and deep local partnerships matter far more than product quality alone.

Ropedia raises US$22M to build data layer for real-world robots: Singapore-based Ropedia secured Series A funding to develop the data infrastructure that enables robots to interpret and navigate unstructured real-world environments, a foundational bottleneck for commercial robotics.

Singapore’s data analysts trust AI to work, not to think: A survey of Singapore-based data professionals finds practitioners are comfortable delegating execution to AI tools but remain deeply reluctant to cede analytical judgement or strategic interpretation.

---

INTERNATIONAL

Travis Kalanick’s robotics firm raises US$1.7B led by a16z: The former Uber CEO’s new venture secured one of the year’s largest robotics rounds, with Andreessen Horowitz leading, a signal that autonomous systems are drawing serious institutional capital again.

SoftBank nears US$40B loan syndication for OpenAI: The Japanese conglomerate is close to finalising a massive credit facility to fund OpenAI’s expansion, underscoring SoftBank’s deepening financial entanglement with the world’s most prominent AI lab.

OpenAI’s AI spending reaches US$750B: OpenAI has disclosed cumulative AI infrastructure expenditure of US$750B, a figure that illustrates the extraordinary capital intensity of frontier model development and the gap widening between top-tier labs and everyone else.

ServiceNow invests US$40M in India’s BusinessNext for APAC banking: The US enterprise software firm’s strategic investment targets autonomous banking deployment across Asia Pacific, with financial institutions in SEA among the primary target markets.

Google justifies AI spending with booming cloud revenue: Alphabet’s latest earnings show Cloud revenue accelerating sharply, giving Google the financial cover to continue heavy AI infrastructure investment, with direct implications for SEA cloud pricing and enterprise adoption.

Tesla’s robotaxi programme stalls amid operational setbacks: Early operational data from Tesla’s robotaxi rollout shows the service underperforming expectations, a cautionary data point for SEA mobility startups tracking autonomous vehicle timelines.

Jack Dorsey launches Buzz to take on Slack with AI agents: Buzz integrates AI agents directly into team communication workflows, positioning it as an agentic-first alternative to Slack as enterprise AI adoption accelerates.

Meta exits major clean energy pact as gas buildout grows: Meta’s withdrawal from a prominent clean energy alliance signals a strategic pivot toward natural gas to power its AI data centres, a tension SEA governments will watch as they negotiate data centre energy commitments.

---

CYBERSECURITY

AI phishing is making trust APAC cybersecurity’s weakest link: Hyper-personalised AI-generated phishing attacks are exploiting the social trust norms prevalent in APAC business culture, making the region disproportionately vulnerable compared with Western markets.

AI already inside the enterprise. Has Asia’s security kept up?: An assessment of enterprise AI adoption in Asia finds security infrastructure lagging significantly behind deployment speed, with shadow AI use creating blind spots in corporate risk management.

OpenAI says Hugging Face was breached via pre-release models: OpenAI disclosed that Hugging Face systems were compromised through access to pre-release models, exposing vulnerabilities in how AI model-sharing platforms manage access controls and pre-deployment security.

---

SEMICONDUCTOR

AI chip startup Etched hits US$10.3B valuation: Etched, which builds chips purpose-built for transformer models, reached a US$10.3B valuation backed by prominent investors, a major vote of confidence in application-specific AI silicon over general-purpose GPU architectures.

AMD secures Anthropic investment to challenge Nvidia in AI chips: The Anthropic-AMD partnership to develop alternative AI training chips is the clearest signal yet that major AI labs are actively funding supply-chain diversification away from Nvidia dependency.

South Korea’s chipmakers eye opportunities amid US-China tensions: Korean semiconductor firms are repositioning to capture market share as US-China chip restrictions create gaps in the global supply chain, with SEA nations emerging as potential beneficiaries of supply chain rerouting.

Singapore’s Tikva Allocell raises US$8M Series A: Tikva Allocell, a Singapore-based biotech developing cell therapy manufacturing technology, secured Series A funding led by Kantharos Capital to scale its proprietary cell processing platform.

---

AI

Anthropic updates Claude voice mode with stronger models: Claude’s upgraded voice mode now runs on more capable underlying models, narrowing the gap with OpenAI’s Advanced Voice Mode and intensifying competition in the real-time conversational AI segment.

Most AI projects fail on workflow, not technology: Organisations rushing to deploy AI without first redesigning underlying processes are setting themselves up for failure — a pattern particularly prevalent among SEA enterprises adopting AI for the first time.

AI-powered automation reshaping SME operations in SEA: Small and medium enterprises across Southeast Asia are deploying AI to automate back-office functions, with early adopters reporting measurable gains in operational efficiency and cost reduction.

How AI is dismantling the risk pool in insurance: AI-driven hyper-personalisation of insurance pricing is eroding the actuarial foundations of traditional risk pooling, with significant implications for insurtech regulation and access to coverage across SEA.

AI empowered teams are shrinking and that’s harder than it sounds: As AI tools enable smaller teams to do more, the organisational and human costs of headcount reduction are proving far more complex than efficiency metrics suggest.

The end of headcount as a success metric: Venture-backed startups and their investors are rethinking how scale is measured as AI enables leaner teams to generate outsized output, challenging a decade of growth-stage hiring orthodoxy.

The real difference between OpenAI and Anthropic: Beyond capability benchmarks, the divergence between OpenAI and Anthropic’s strategies becomes starkest when AI becomes a commodity — one built for scale, the other for trust.

Taiwan’s stablecoin moment: NTD could outshine the dollar: A case for Taiwan dollar-backed stablecoins argues that Taiwan’s trade surplus, currency stability, and semiconductor-driven export economy make it a compelling anchor for a non-USD stablecoin, relevant as SEA explores digital currency infrastructure.

---

THOUGHT LEADERSHIP

SEA startups can no longer sit on the fence on tech stack: Geopolitical fragmentation is forcing SEA founders to choose sides on cloud, chip, and software infrastructure, a decision with long-term strategic consequences they can no longer defer.

Southeast Asia as a climate capital proving ground: SEA is emerging as the test bed for climate finance models that blend blended finance, carbon markets, and impact investing, a convergence that could attract a new class of global capital.

Thought leadership as a market entry tool: Founders expanding beyond their first market can use strategic content and public positioning to build trust and brand credibility in new geographies before sales teams arrive.

The most sophisticated AI strategy is a puzzle hunt in Toa Payoh: Singapore’s grassroots AI adoption story, told through a community scavenger hunt, argues that the most durable AI strategies are built bottom-up, not imposed top-down by strategy consultants.

Multi-channel marketing is survival, not strategy, in Asia: Single-channel dependency is a growth ceiling for Asian brands. This piece makes the case that diversified marketing is now a baseline operational requirement, not a competitive differentiator.

Why remote no longer means anywhere: The era of frictionless remote work is ending as employers tighten location requirements and compliance burdens grow, with real consequences for SEA’s talent mobility and hiring models.

Product management as method acting: The best PMs don’t just research users; they internalise them, adopting their mental models and constraints to build products that solve real problems rather than assumed ones.

The post Ecosystem Roundup: When AI acts, humans are still responsible appeared first on e27.

For years, cell therapy has carried one of oncology’s most striking contradictions. It has changed the outlook for some blood cancer patients, yet has struggled to make the same impact in solid tumours, which account for the vast majority of cancer cases worldwide.

Singapore-based Tikva Allocell is trying to push through that wall with a different kind of off-the-shelf cell therapy. The biotechnology company has raised US$8 million in Series A financing led by Kantharos Capital, with the proceeds earmarked for studies needed before human testing and a planned Investigational New Drug, or IND, submission by the end of 2026.

Also Read: Singapore’s Biobot Surgical raises US$15.6M to take prostate-care robot global

If regulators clear the application, Tikva plans to begin a Phase 1 clinical trial of its lead candidate, TAVST01, in patients with advanced B7-H3-positive cancers at sites in Singapore and the US.

That dual geography matters. For Singapore, which has spent years building its biomedical research base, the trial would place a homegrown cell therapy company on a path that connects local clinical infrastructure with the world’s largest biotech market. For patients, the more important question is whether Tikva’s approach can solve a problem that has repeatedly defeated the field: how to make donor-derived immune cells survive long enough inside a patient to attack solid tumours.

A different starting point for cell therapy

TAVST01 targets B7-H3, a protein found across several difficult-to-treat solid tumours, including lung, breast, prostate, pancreatic and paediatric cancers. B7-H3 has attracted interest because it is often highly expressed on cancer cells and in the tumour microenvironment, while its presence in normal tissues appears more limited, making it a potential target for cancer therapies.

The company’s approach begins not with a generic donor T cell, but with Epstein-Barr virus-specific T cells. Epstein-Barr virus, or EBV, is extremely common; most adults carry it from a past infection, and the immune system typically keeps a long-lived population of EBV-fighting T cells on patrol. Tikva’s bet is that these cells may offer the durability that conventional donor-derived cell therapies have lacked.

“Cell therapy has transformed the treatment of blood cancers but has repeatedly stalled at the solid-tumour door; the donor cells either fail to persist or are eliminated by the patient’s immune system before they can act,” said Ivan Horak, founder and CEO of Tikva Allocell.

“We started from a different place: a virus-fighting T cell the body naturally sustains, armed to seek out B7-H3 and engineered to withstand the rejection that defeats most donor-derived approaches, with minimal gene editing,” he added.

Also Read: AI is detecting cancer earlier in Southeast Asia but our policies and capital have not caught up

Tikva’s platform, called ALLO SerpinB9 EBVST, is licensed exclusively from Baylor College of Medicine and further enhanced through the company’s own protein-engineering work. The cells are fitted with a B7-H3-targeting receptor and an optimised version of SerpinB9, a naturally occurring inhibitor of granzyme B. Granzyme B is one of the enzymes immune cells use to kill their targets.

In simple terms, a patient’s immune system would normally recognise donor cells as foreign and attack them. Tikva’s SerpinB9 “armour” is designed to help the therapy resist that attack, remain active for longer, and reduce one of the central weaknesses of allogeneic, or donor-derived, cell therapy. The company also says the approach is designed to minimise graft-versus-host disease, a serious complication in which donor immune cells attack the patient’s healthy tissues, while requiring only limited gene editing.

Why solid tumours remain hard

The promise of cell therapy is best known through CAR-T treatments, where a patient’s immune cells are engineered to recognise cancer and then infused back into the body. These therapies have delivered strong results in some blood cancers, but solid tumours are a different battlefield.

Tumour masses are physically harder for immune cells to penetrate. They often create an immunosuppressive microenvironment, a local shield that weakens immune attacks. Antigens, the markers therapies use to identify cancer cells, can vary across tumour cells, allowing some cancer cells to escape. And when cells come from a donor rather than the patient, the recipient’s immune system may quickly eliminate them.

Allogeneic therapies are attractive because they can be manufactured in advance, stored, and potentially given to many patients without waiting weeks for a bespoke treatment. That could make them cheaper, faster and more scalable than patient-specific therapies. But the trade-off has been persistence: if the donor cells disappear too quickly, they may not have time to do meaningful work.

Tikva’s answer is to use a type of immune cell the body is already used to maintaining, then engineer it to both recognise B7-H3 and resist immune rejection. In preclinical work, the company says TAVST01 has shown potential to kill tumour cells directly and to remodel the tumour microenvironment that has held back other solid-tumour cell therapy attempts.

The next step is more demanding. IND-enabling studies will test whether the therapy is safe enough, consistent enough and well-characterised enough for regulators to allow human trials. For a young biotech, this stage is capital-intensive and unforgiving, which makes the Series A round central to Tikva’s timetable.

A Singapore biotech with global ambitions

Tikva’s financing also reflects a broader shift in Southeast Asia’s life sciences ecosystem. The region is better known in tech circles for fintech, e-commerce and logistics startups, but Singapore has long treated biomedical science as a strategic sector, supported by research institutes, hospital networks, manufacturing capacity and regulatory infrastructure.

Still, building a biotech company in Southeast Asia is very different from building a software startup. Timelines are longer, capital requirements are heavier, and the path to revenue usually runs through clinical data, regulatory approval and partnerships with larger pharmaceutical companies. For Singapore-based biotechs, the challenge is not only to do credible science, but to connect early research with global clinical and commercial pathways.

Tikva appears to be structuring itself with that in mind. By planning clinical sites in both Singapore and the US, it can anchor development in its home market while engaging the regulatory and clinical ecosystem that often determines whether biotech assets attract global investors, partners or acquirers.

“Our investment reflects strong conviction in both Tikva’s science and its leadership team,” said Terence Tan, Managing Partner at Kantharos Capital. “Tikva is addressing fundamental challenges that have constrained allogeneic cell therapies, and we believe its ALLO SerpinB9 EBVST platform can extend the reach of cell therapy to solid-tumour patients who today have limited options.”

The competitive field

Tikva is entering a crowded and technically difficult race. Globally, companies such as Fate Therapeutics, Allogene Therapeutics, Caribou Biosciences, and Atara Biotherapeutics have explored allogeneic cell therapies, while larger pharmaceutical and biotech players continue to invest in CAR-T, T-cell receptor therapies and natural killer cell platforms. In solid tumours, B7-H3 is also being pursued through different modalities, including antibody-drug conjugates, bispecific antibodies and cell therapies. That means Tikva will not be judged on novelty alone. It will need to show that its EBV-specific, SerpinB9-armoured cells can persist, avoid serious safety issues, and generate signals of tumour activity in patients who have few remaining options.

Also Read: Danish startup Alcolase is betting its future on Asia’s drinking culture

For now, the company’s story remains preclinical. The US$8 million round does not prove that TAVST01 will work in humans, nor does it remove the biological risks that have humbled many solid-tumour programmes before it. But it does give Tikva enough runway to test a clear hypothesis: that a naturally persistent virus-specific T cell, properly engineered, can become a practical off-the-shelf weapon against solid cancers.

If that hypothesis survives clinical testing, the implications would reach well beyond one Singapore startup. It would strengthen Southeast Asia’s claim to a place in high-end therapeutic innovation, not merely as a trial site or manufacturing base, but as a source of globally relevant biotech platforms.

The post Singapore’s Tikva targets solid cancer barrier with US$8M Series A appeared first on e27.